This was the article I was thinking of, per my earlier post, not the
Jeff Sonas article.

It seems likely that whatever conclusions may be
drawn from such studies are largely determined by
the way in which the study is constructed, which
is just the opposite of what is supposedly desired
(i.e. computer-like objectivity).

For instance, had the study shown "desirable"
results right off the bat, the need to compensate
for the simplicity of position would never have
even occurred. If it turned out that, say, GM
Capablanca was more accurate because he
preferred simple positions, this could have been
interpreted as meaning he was simply the
strongest player; instead, there arose an
"emotional need" to compensate for some
assumed flaw, as if his choice of style were
somehow unfair to the other contenders.

What is never shown and rarely mentioned
is all the tweaking of the various formulae
that goes on before finalizing the charts and
results presented to us as readers, and this
invisible stuff is precisely what determines
the final rankings.

-- help bot